MAL Simulator for Automated Cyber Attack and Defense Agents
September 16, 2026
The MAL Simulator uses the Meta Attack Language to model cyber operations for decision-driven simulation. Researchers trained reinforcement learning agents that outperformed search-based methods for attackers and lowered costs compared to heuristic agents for defenders in emulated network environments.
HOW THIS AFFECTS YOU
●
builderThis provides a simulation environment for testing automated cyber security agents without live network risk.
●
researcherYou can use this framework to train and evaluate RL-based offensive and defensive agents.